Go Back   nV News Forums > Linux Support Forums > NVIDIA Linux

Newegg Daily Deals

Reply
 
Thread Tools
Old 01-20-11, 11:26 AM   #25
jpi110
Gentoo User
 
Join Date: Jan 2011
Location: Portland, Oregon
Posts: 14
Default Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

Quote:
Originally Posted by oyvind View Post
I tried using the latest 260.19.29 driver for a while, to see if things were in fact stable, even though scary errors are spewed out into my kernel log at times. Suspend/resume works, and the system looks stable on the surface.. .. Except for the "gray screen of death" that just hit me. Out of nowhere (when editing in Emacs), system froze and screen turned completely ## GRAY ##. Alt+SysRq dead, network dead, everything dead. So it is not usable nor stable in practice, which would also be the logical conclusion judging by the look of the error messages that appear in the kernel log. It was worth a shot, I guess ..

Please consider releasing something that works for the NVS 3100M and that is as stable as 195.36-series. Because 195.36.24 is solid (I've had uptime for more than a month using this driver, with lots of suspend/resume and hibernate cycles). If you need more information, logs or debugging data, please let me know, I'd be glad to help in any way I can ..
Unfortunately, this seems off topic. If you are not seeing the error messages listed in the subject of this thread, then you should follow the forum rules about creating a new thread with the pertinent requirements. See the stickies in this forum for information.
jpi110 is offline   Reply With Quote
Old 01-20-11, 01:51 PM   #26
oyvind
Registered User
 
Join Date: May 2004
Location: Norway
Posts: 117
Default Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

Quote:
Originally Posted by jpi110 View Post
Unfortunately, this seems off topic. If you are not seeing the error messages listed in the subject of this thread, then you should follow the forum rules about creating a new thread with the pertinent requirements. See the stickies in this forum for information.
Please read the entire thread. I'm the OP, I started the thread and posted all the original errors and logs, and the stability issues I mention here are very much related.
oyvind is offline   Reply With Quote
Old 01-23-11, 07:43 PM   #27
oyvind
Registered User
 
Join Date: May 2004
Location: Norway
Posts: 117
Thumbs down Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

Still an issue with latest 270.18 beta driver.

Code:
$ dmesg|grep NVRM
[    5.265666] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  270.18  Tue Jan 18 21:46:26 PST 2011
[   18.470535] NVRM: os_raise_smp_barrier(), invalid context!
[   18.485152] NVRM: os_raise_smp_barrier(), invalid context!
[  101.598273] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
It's as unstable as ever, no improvements over earlier releases. And now the "os_raise_smp_barrier(), invalid context!" messages are back as well. Those were not present with 260.19.29 (at least I did not observe them even once during testing, but have seen them in earlier releases).

Again, I urge you to please look into this, or at least provide some form of feedback (e.g. if this is not a nvidia problem and Dell has fscked up their hardware or something in the Linux kernel is b0rked and there's nothing you can do.).

Attaching updated bug report log using driver 270.18.
Attached Files
File Type: gz nvidia-bug-report.log.gz (41.3 KB, 70 views)
oyvind is offline   Reply With Quote
Old 01-23-11, 10:01 PM   #28
oyvind
Registered User
 
Join Date: May 2004
Location: Norway
Posts: 117
Default Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

OK, se here's another clue: If I disable "Sync to vblank" from within Compiz, then the "os_schedule: Attempted to yield the CPU while in atomic"-messages seem to vanish and console-switching becomes much more reliable. The "os_raise_smp_barrier()" messages still persist, though.

Disabling this Compiz-option introduces tearing in rendering, even though OGL vblank is enabled in nvidia-settings.
oyvind is offline   Reply With Quote
Old 01-25-11, 09:29 AM   #29
voegel
Registered User
 
Join Date: Dec 2010
Posts: 12
Default Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

I upgraded the nvidia driver to 260.19.36. No change.

The hint relating to "sync to vblank" is very interesting. I will test that in nvidia-setting and the application itself (in my case: mythtv).
voegel is offline   Reply With Quote
Old 01-25-11, 10:36 AM   #30
voegel
Registered User
 
Join Date: Dec 2010
Posts: 12
Default Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

I just tried disabling OpenGL sync to vblank in nvidia-settings with no effect. Disabling in the application isn't possible. I wasn't aware that this option has been removed with mythtv 0.24.

An upgrade to 270.18 didn't help either.
voegel is offline   Reply With Quote
Old 01-26-11, 09:15 AM   #31
oyvind
Registered User
 
Join Date: May 2004
Location: Norway
Posts: 117
Default Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

Quote:
Originally Posted by voegel View Post
I just tried disabling OpenGL sync to vblank in nvidia-settings with no effect. Disabling in the application isn't possible. I wasn't aware that this option has been removed with mythtv 0.24.

An upgrade to 270.18 didn't help either.
If you read a bit more carefully, you'll see that it's the "Sync to vblank" option in Compiz-settings that I'm talking about (ccsm), not the "opengl vblank sync" option in nvidia-settings.

Anyway, after disabling vblank-sync in Compiz, everything has been stable, including games and suspend/resume. I still get oodles of "os_raise_smp_barrier()" in the kernel log though, and that isn't exactly pretty. I'm running everything with "__GL_SYNC_TO_VBLANK" env var set, so OpenGL-apps sync good enough (though Compiz is not 100% tear free). This variable should be same setting that is present in nvidia-settings.
oyvind is offline   Reply With Quote
Old 01-31-11, 12:44 PM   #32
voegel
Registered User
 
Join Date: Dec 2010
Posts: 12
Default Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

Quote:
Originally Posted by oyvind View Post
If you read a bit more carefully, you'll see that it's the "Sync to vblank" option in Compiz-settings that I'm talking about (ccsm), not the "opengl vblank sync" option in nvidia-settings.
I have completely understood that point. But as I mentioned, I cannot control this feature in mythtv as of version 0.24 any more. This option has been removed.

Quote:
Originally Posted by oyvind View Post
Anyway, after disabling vblank-sync in Compiz, everything has been stable, including games and suspend/resume. I still get oodles of "os_raise_smp_barrier()" in the kernel log though, and that isn't exactly pretty. I'm running everything with "__GL_SYNC_TO_VBLANK" env var set, so OpenGL-apps sync good enough (though Compiz is not 100% tear free). This variable should be same setting that is present in nvidia-settings.
In my case I just have the "NVRM: os_schedule: Attempted to yield the CPU while in atomic[...]" error message. The "os_raise_smp_barrier()" problem, I don't experience.
voegel is offline   Reply With Quote

Old 02-04-11, 02:20 PM   #33
voegel
Registered User
 
Join Date: Dec 2010
Posts: 12
Default Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

I seem to have found the issue. Unfortunately it only seems as if the graphics card driver is the problem. It looks as if it is "x11vnc" in combination with xserver 1.9.0.

There is a thread to solve the problem here: http://ubuntu-ky.ubuntuforums.org/sh...1612704&page=3

The reason why it looks like an Nvidia problem is, because the is no crash report. And the "NVRM: os_schedule: Attempted to yield the CPU while in atomic[...]" message is always printed after a start of the Xserver. And since it ends up in login screen - X11 doesn't get startet then any more - it is the last entry in the log.

To make it final I will test this now. But it looks good so far.
voegel is offline   Reply With Quote
Old 02-05-11, 06:00 AM   #34
oyvind
Registered User
 
Join Date: May 2004
Location: Norway
Posts: 117
Default Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

Quote:
Originally Posted by voegel View Post
I seem to have found the issue. Unfortunately it only seems as if the graphics card driver is the problem. It looks as if it is "x11vnc" in combination with xserver 1.9.0.

There is a thread to solve the problem here: http://ubuntu-ky.ubuntuforums.org/sh...1612704&page=3

The reason why it looks like an Nvidia problem is, because the is no crash report. And the "NVRM: os_schedule: Attempted to yield the CPU while in atomic[...]" message is always printed after a start of the Xserver. And since it ends up in login screen - X11 doesn't get startet then any more - it is the last entry in the log.

To make it final I will test this now. But it looks good so far.
I think it's a more general nvidia driver problem which is triggered by different things. I get this error a lot on my Dell laptop with the newer drivers, but I don't use any VNC server. I remember though that nvidia did fix a bug related to screen grabbing and Compiz, which is a difference between 195.X and newer drivers (don't remember exactly in which version this was fixed, i.e. Compiz would grind to a halt when using VNC to access desktop).
oyvind is offline   Reply With Quote
Old 02-06-11, 07:32 AM   #35
voegel
Registered User
 
Join Date: Dec 2010
Posts: 12
Default Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

Everything seems to be stable so far. I do still get this message on startup of X. But only on startup, not during the session. Which was the case also before.

So in my situation, the problem clearly wasn't the nvidia driver but rather the x11vnc.
voegel is offline   Reply With Quote
Old 02-15-11, 01:46 PM   #36
oyvind
Registered User
 
Join Date: May 2004
Location: Norway
Posts: 117
Default Re: NVRM: os_schedule: Attempted to yield the CPU while in atomic[...] [260.19.21 x86

Still not fixed in latest 270.26 beta. Error occurs when Compiz VSYNC is enabled and I switch to console or use suspend. X freezes/hangs on and off afterwords (e.g. acts weird). Seems like disabling Compiz VSYNC still works around the issue.

Code:
$ dmesg|grep NVRM
[    3.500884] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  270.26  Sun Feb 13 22:30:06 PST 2011
[   66.246885] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
[  121.537777] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  122.938731] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  122.951252] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  122.951435] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  122.963807] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  122.963942] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  122.990852] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  123.207755] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  123.217513] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  123.252276] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  123.311589] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  123.365436] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  123.576265] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  123.763055] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  124.362794] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  124.629250] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00000005
[  279.434594] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
[  424.332137] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context

The "NVRM: os_raise_smp_barrier(), invalid context!" messages seem to be gone now, in contrast to 270.18, which is good I guess..

Nvidia, got any updates on this issues, any info at all ? Attaching updated bug report log.
Attached Files
File Type: gz nvidia-bug-report.log.gz (52.5 KB, 67 views)
oyvind is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT -5. The time now is 09:57 PM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.